RC 90 - occur only for partitions of not active cluster system node

[ziro]

Hello,

RC 90 still appear only for disks partitions which are not visible to not active cluster node. So fullbackup of non active node make a backup of C:\ D:\ E:\ and Shadow Copy Components but for F:\ G:\ H:\ I:\ J:\ K:\ L:\ still occur RC 90.

Backup server and client run on version 7.1.0.3

We tried to use command:

w2koption -backup -ignore_unresolved_volumes F:G:H:I:J:K:L:

and restart NetBackup on client, it didnt solve this issue.

Client's server is restarted on daily basis and RC 90 for fullbackups occur everyday.

[RamNagalla]

it looks like you are using the Multi-stream backup with indivudial backup selection for each drive.

1) remove the Multi-stream backup

 or 

2)add All_Local_Drives in backup selection instead of each drive.

hope this helps.

[RLeon]

Please refer to this technote for how to properly backup failover cluster shared resources (still applies to Nbu 7.x):
http://www.symantec.com/business/support/index?page=content&id=TECH35842

If your cluster nodes are nbu clients and not media servers, then just follow the part starting from:

3. NetBackup as Network Client

Basically, non clustered/shared local resources (in your case C:\ D:\ E:\) on each node should be in their own policy(ies) using their real hostnames.

For the clustered/shared resources (in your case F:\ G:\ H:\ I:\ J:\ K:\ L:\) they have to be in a separate policy that uses the cluster virtual server name.

[ziro]

Thanks for hints. You both are right.

But I cant remove Multi-streams because huge backup of all partitions will run more than 3 days.

There is already set up ALL_LOCAL_DRIVES as backup selection.

We cant use cluster name of server (only nodes separetely), cluster server has not own IP of backup LAN so in our case its not possible to set it up this way.

I think that problem is that cluster with both nodes is restarted everyday at the same time and at this time both nodes check all partitions, each node register all partitions, and after that are some partitions (F:\ G:\ H:\ I:\ J:\ K:\ L:\) switched to active node. But each node thinks that all partitions belong to him, so netbackup trying to backup them all without success and RC90 appear.

What do you think about this theory?

[Marianne]

See this TN: http://www.symantec.com/docs/TECH173606

MSCS passive node backup fails with error 90

... cluster server has not own IP of backup LAN

Surely this is easy enough to put in place?

[RLeon]

But each node thinks that all partitions belong to him, so netbackup trying to backup them all without success and RC90 appear.

What do you think about this theory?

The nodes don't both "think" they own the partitions. If they do, your data would have been corrupted.

It is simply that your NetBackup configuration is not correctly setup for such clustered resources, as documented in both technotes. You can use multi stream, but you cannot use "ALL_LOCAL_DRIVES" with the real node hostnames and the virtual server name in such a case.

If your cluster does not have a virtual server name with a virtual server ip that is associated with your shared resources, it is not a correctly setup failover cluster configuration. If you are using a backup lan, the virutal server name and ip will also need their backup lan counterpart setup for netbackup to use, so that backup traffic will be correctly routed though the backup lan.

[RamNagalla]

hi,

if you can not remove the Mulitstream and also can not configure the backup with Cluster IP.

I have a workaround for your issue, but it requires you to create Multiple policies.

create a seperate policy for each Cluster Drive and add one  C:/Program Files/Veritas/Netbackup/( any dummy file or folder) also to the backup backup selection with out multi-streaming

for policy backup selection is looks like, below with out Mulitstreaming.

F:\

C:/Program Files/Veritas/Netbackup/

and create separeate policy for each Drive as above. and test the backups.

as you are creating seperate policy for each Drive, its just like Mulitsteaming, 

hope this helps.

[ziro]

Hello,

Yes, your hints and workaround will solve this issue. Thanks.

I just realize that main problem is that I cant configure it that way because of rules and standardization processes on our backup server :(

I cant create backup policy for each node of cluster and do backup of active partitions because visibility of partitions which are active is changing. And I'm not admin of these client's servers(nodes) so I dont know which partitions are active, when new partition is added or deleted, I dont have info about it. I also cant create new backup policy for one server, we use minimal count of policies which are connected to our monitoring tool and its lot of work to implement new policy to this monitorng tool. And also its not feasible to backup cluster server, only each node of cluster, because cluster server cannot be inplemented to our asset and monitoring tools.

There is no possibility to set it up via cluster name server, this server has not backup LAN, only nodes have own backup LAN with NIC used only for backups. So I wondering and seeing possibility to ensure backup LAN also for cluster server. But its lot of negotiations with other responsible teams. I hope it will be possible to set it up via cluster server and it solve this issue.

Thanks a lot for your efforts.

[RLeon]

The best way to monitor NetBackup is to use OpsCenter. It's the official tool and it's free. In most cases it should provide all your NetBackup monitoring needs, such as customisable e-mail alerts. A paid version is also available with more features.

NetBackup 7 OpsCenter and OpsCenter Analytics
http://eval.symantec.com/mktginfo/enterprise/white_papers/b-nbu_7_opscenter_analytics_WP.en-us.pdf

Symantec OpsCenter 7.5 Administrator's Guide
http://www.symantec.com/business/support/index?page=content&id=DOC5042

Try not to let misconfigurations propagate and filter down and across your environment. I.e., if a failover cluster is not configured properly, then you have to use improper backup settings to protect it, which then requires non standard and arcane methods for monitoring, and so on. When it comes the time to review/audit/upgrade/fine tune the whole lot, it'll be..... refreshing.

[Marianne]

There is no possibility to set it up via cluster name server, this server has not backup LAN, only nodes have own backup LAN with NIC used only for backups.

Work with your cluster/sysadmin.

Both nodes already have NICs on backup lan.
All that is required is to config a virtual IP address on backup lan as part of the cluster config.